Reversible Photoinduced Reductive Elimination of H2 from the

Jan 20, 2016 - Reversible Photoinduced Reductive Elimination of H2 from the Nitrogenase ...... Dance , I. Chem. ...... Dust accidents top CSB agendaï»...
2 downloads 0 Views 1MB Size
Article pubs.acs.org/JACS

Reversible Photoinduced Reductive Elimination of H2 from the Nitrogenase Dihydride State, the E4(4H) Janus Intermediate Dmitriy Lukoyanov,† Nimesh Khadka,§ Zhi-Yong Yang,§ Dennis R. Dean,# Lance C. Seefeldt,*,§ and Brian M. Hoffman*,† †

Department of Chemistry, Northwestern University, Evanston, Illinois 60208, United States Department of Chemistry and Biochemistry, Utah State University, Logan, Utah 84322, United States # Department of Biochemistry, Virginia Tech, Blacksburg, Virginia 24061, United States §

S Supporting Information *

ABSTRACT: We recently demonstrated that N2 reduction by nitrogenase involves the obligatory release of one H2 per N2 reduced. These studies focus on the E4(4H) “Janus intermediate”, which has accumulated four reducing equivalents as two [Fe-H-Fe] bridging hydrides. E4(4H) is poised to bind and reduce N2 through reductive elimination (re) of the two hydrides as H2, coupled to the binding/reduction of N2. To obtain atomic-level details of the re activation process, we carried out in situ 450 nm photolysis of E4(4H) in an EPR cavity at temperatures below 20 K. ENDOR and EPR measurements show that photolysis generates a new FeMo-co state, denoted E4(2H)*, through the photoinduced re of the two bridging hydrides of E4(4H) as H2. During cryoannealing at temperatures above 175 K, E4(2H)* reverts to E4(4H) through the oxidative addition (oa) of the H2. The photolysis quantum yield is temperature invariant at liquid helium temperatures and shows a rather large kinetic isotope effect, KIE = 10. These observations imply that photoinduced release of H2 involves a barrier to the combination of the two nascent H atoms, in contrast to a barrierless process for monometallic inorganic complexes, and further suggest that H2 formation involves nuclear tunneling through that barrier. The oa recombination of E4(2H)* with the liberated H2 offers compelling evidence for the Janus intermediate as the point at which H2 is necessarily lost during N2 reduction; this mechanistically coupled loss must be gated by N2 addition that drives the re/oa equilibrium toward reductive elimination of H2 with N2 binding/reduction.



INTRODUCTION Biological nitrogen fixationthe reduction of N2 to two NH3 moleculesis primarily catalyzed by the Mo-dependent nitrogenase. This enzyme comprises two component proteins, denoted the Fe protein and the MoFe protein. The former delivers electrons one at a time to the MoFe protein, where they are utilized at the active-site iron−molybdenum cofactor ([7Fe-9S-Mo-C-R-homocitrate]; FeMo-co, Figure 1) to reduce substrate.1,2 Kinetic studies of N2 reduction by nitrogenase, carried out in the 1970s and 1980s by many groups, especially by Lowe and Thorneley and their co-workers, culminated in the Lowe−Thorneley (LT) kinetic model for nitrogenase function.1,3,4 It describes the kinetics of transformations among catalytic intermediates, denoted En, where n is the number of electron/proton deliveries to the catalytic FeMo-co, with electron transfer from the partner Fe protein in each of these steps being driven by the binding and hydrolysis of two MgATPs within the Fe protein.5 A central defining feature of this scheme is a mysterious and puzzling, obligatory (mechanistic) requirement for the formation of one H2 for each N2 reduced. This, in turn, leads to a limiting eight-electron enzymatic stoichiometry for enzyme-catalyzed nitrogen fixation, eq 1, a conclusion in agreement with stoichiometric experiments by Simpson and Burris.6 However, the obligatory requirement for H2 formation has not been universally accepted.7 Most tellingly, in their © XXXX American Chemical Society

Figure 1. Crystal structure of FeMo-co. Fe is shown in rust, Mo in magenta, S in yellow, carbide in dark gray, C in gray, N in blue, and O in red. The Fe-atoms of the catalytic 4Fe-4S face are labeled as 2, 3, 6, and 7. Two amino acids, α-70Val and α-195His, around the FeMo-co are also shown; either just the former or both are modified in the enzyme used in this study (see text). The image was created using coordinates for PDB 2AFI.

magisterial review, Burgess and Lowe themselves questioned this requirement:1 “Thus, the data that support the obligatory evolution of one H2 for every N2 reduced are much less compelling than the data that require us to believe that some H2 will always be evolved during N2 reduction.” Received: November 6, 2015

A

DOI: 10.1021/jacs.5b11650 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

limiting stoichiometry of eight electrons/protons for the reduction of N2 to two NH3 (eq 1).16 But these efforts, while establishing the re/oa mechanism for nitrogenase activation, do not provide atomic-level details of the re activation process. For example, in rough analogy to nucleophilic substitution in organic chemistry, we can imagine a spectrum of reaction pathways for re/oa, as illustrated in Figure 3: Is

N2 + 8e− + 16ATP + 8H+ → 2NH3 + H 2 + 16ADP + 16Pi

(1)

We recently proposed8,9 that obligatory H2 formation was required to explain the multitude of mechanistic observations by numerous investigators that had accumulated over decades.1 This proposal focuses on the E4(4H) “Janus intermediate” (see Figure 2 for notation), which has accumulated four of the eight

Figure 2. Schematic of re/oa equilibrium. The cartoon represents the Fe 2, 3, 6, and 7 faces of FeMo-co, and “2N2H” implies a species at the diazene reduction level of unknown structure and coordination geometry. In the indicated equilibrium, the binding and activation of N2 are mechanistically coupled to the re of H2, as described in the text. In the En notation, n = number of e−/H+ added to FeMo-co; parentheses denotes the stoichiometry of H/N bound to FeMo-co.

Figure 3. Schematic of alternative limiting mechanisms for re/oa equilibrium.

the conversion concerted? Associative? Dissociative? If a discrete intermediate (e.g., Y, or Z) exists, what are its properties? To obtain atomic-level details of the re activation process requires deeper understanding of the inorganic chemistry of the bridging hydrides of E4(4H): What are their properties, and what properties do they confer on this enzyme state? We here present the initial results of a photochemical approach inspired by the properties of inorganic dihydride complexes. The photolysis of transition metal dihydride complexes with mutually cis hydride ligands commonly results in the release of H2.17−26 As noted by Perutz,17 “The photochemical reaction causes a reduction in oxidation state of two and is a typical example of reductive elimination. The reverse reaction will usually proceed thermally and is the prototype example of an oxidative addition reaction.” Regardless of the precise nature of the thermal re/oa equilibrium process in nitrogenase (Figure 2), photoinduced re would cleanly give an activated version of the doubly reduced E2(2H) intermediate, which we denote E4(2H)*, that would be analogous to the intermediate that would form upon thermal dissociative re loss of H2 prior to N2 binding (Z, Figure 3). Our cartoon depictions of E4(4H) frequently have shown the two [Fe−H−Fe] hydrides with a common vertex (see Chart 1) in order to emphasize the analogy between re of H2 by

required reducing equivalents, storing them as two [Fe-H-Fe] bridging hydrides.10−12 E4(4H) sits at a transition in the N2 reduction pathway, poised to “fall back” to E0 by release of two H2, but equally poised to bind and reduce N2 through the accumulation of four more equivalents, hence the appellation.9 The bridging mode of hydride binding plays a key mechanistic role. Bridging hydrides are less susceptible to protonation than terminal hydrides, and so they diminish the tendency of FeMo-co to “fall back” by losing reducing equivalents through the formation of H2. However, the bridging mode also lowers hydride reactivity relative to that of terminal hydrides.13,14 How this “deactivated” intermediate becomes activated through the release of H2 coupled to N2 binding forms part of the “mystery” of dinitrogen fixation by nitrogenase. We proposed that the E4(4H) state becomes activated for the binding of N2 and its hydrogenation to a N2H2-level moiety through the reductive elimination (re) of the two hydrides as H2, the forward direction of the equilibrium in Figure 2.8,9 This proposal was initially supported8,9 by showing that the behavior of nitrogenase during the reverse of this equilibrium, the oxidative addition (oa) of H2 with loss of N2, explains the key constraints on nitrogenase mechanism that had been revealed over the years.1 In particular it explains the previously baffling observation that D2 can only react with nitrogenase during turnover with N2 present, and then is stoichiometrically reduced to two HD.1 Promptly thereafter we confirmed the mechanistic prediction that during turnover under N2/D2, the reverse of the equilibrium of Figure 2, the oa of D2 by the E4(2N2H) intermediate with the loss of N2, must generate the E4(2D2H) isotopologue with D2 having been converted selectively into two bridging deuterides, a state which could form in no other way.15 This observation established the re/oa equilibrium is thermodynamically reversible. More recently, we demonstrated that the (re/oa) activation equilibrium of Figure 2 is not only thermodynamically, but also kinetically reversible. The overall result of these several findings is to establish the mechanistic requirement for the formation of one H2 per N2. This in turn implies the

Chart 1

mononuclear metal dihydrides, and the activation of FeMo-co through re of H2. However, we do not yet know their exact disposition, and know of no precedent for re, either thermal or photochemical, for the “parallel” hydrides drawn in the cartoon (Figure 2), a geometry that is suggested by preliminary DFT computations.27 On the other hand, no inorganic multimetallic dihydride of which we are aware exhibits a 4Fe “face”, as does B

DOI: 10.1021/jacs.5b11650 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society



FeMo-co, and therefore none could have the adjacent, parallel hydrides drawn in the cartoon. Thus, photolysis of the Janus E4(4H) intermediate embeds FeMo-co even more deeply within the body of organometallic chemistry, yet breaks new ground.



Article

RESULTS AND DISCUSSION

Figures 4 and S1 present X-band and Figure S2 presents Q-band EPR spectra of the S = 1/2 E4(4H) intermediate

MATERIALS AND METHODS

Materials and Protein Purifications. All the reagents were obtained from SigmaAldrich (St. Louis, MO) or Fisher Scientific (Fair Lawn, NJ) and were used without further purification. Argon, N2, and acetylene gases were purchased from Air Liquide America Specialty Gases LLC (Plumsteadville, PA). Remodeling the active site of MoFe protein by the α-70Val→Ile mutation permits the freeze trapping of MoFe with high populations of E4(4H).10 Experiments were carried out both with the singly substituted, α-70Ile MoFe protein and with the doubly substituted α-70Val→Ile/α-195His→Gln MoFe proteins. As shown in Table S1, this protein functions similarly to the single mutant. The two proteins were obtained from the corresponding Azotobacter vinelandii strains. They were grown, and the corresponding nitrogenase MoFe proteins were expressed and purified as described elsewhere.28 The handling of all buffers and proteins were done anaerobically under Ar atmosphere or under Schlenk vacuum line unless stated otherwise. EPR and ENDOR Samples. The E4(4H) intermediate and its deuterated analogue, E4(4D), are prepared by turnover of the MoFe protein in H2O and D2O buffers, respectively. Depending on the buffer used, all exchangeable sites are thereby populated with H or D, not only the hydride bridges. Thus, measured kinetic isotope effects associated with re and oa steps of the equilibrium of Figure 229,30 are a composite of primary isotope effects associated with hydride re or H2 oa plus any, smaller, solvent isotope effects. EPR samples were prepared in a dioxygen free buffer containing a MgATP regeneration system with final concentrations of 13 mM ATP, 15 mM MgCl2, 20 mM phosphocreatine, 2.0 mg/mL bovine serum albumin, and 0.3 mg/mL phosphocreatine kinase in a 200 mM MOPS buffer at pH 7.3 (H2O buffer) or pD 7.3 (D2O buffer; pH meter reading of 6.9)31 with 50 mM dithionite. MoFe protein was added at ∼50 μM final concentration and the reaction was initiated by addition of Fe protein at ∼36 μM concentration. The reaction was allowed to run at room temperature under Ar atmosphere for 20−25 s before freeze-quenching the samples. The samples for ENDOR experiments were prepared similarly, but typically with 3-fold higher concentrations. EPR and ENDOR Measurements. X-band EPR spectra were recorded on a Bruker ESP 300 spectrometer equipped with an Oxford Instruments ESR 900 continuous He flow cryostat. To allow illumination, a Bruker ER4104R cavity was employed. This cavity allows front-face optical access through a waveguide beyond cutoff (microwave non-transmitting) on the cavity front face, with a 4 × 10 mm optical transmission path. In situ photolysis of a sample held within the cryostat at the chosen temperature initially employed a 15 mW blue LED inserted into this illumination port. Subsequently it employed a Thorlabs Inc. (Newton, NJ) PL450B, 450 nm, 80 mW Osram Laser Diode mounted on the port through use of the corresponding diode mount with focusing lenses. Thermal relaxation was monitored by the step-annealing procedure in which the sample was quickly warmed to a desired temperature, held there for a fixed time, then promptly returned to 77 K, and then examined by EPR at a still lower temperature. Q-band CW EPR and 1,2H ENDOR spectra were collected on a spectrometer with a helium immersion Dewar as previously reported.32 The stochastic field-modulation detected ENDOR technique, first reported by Brueggeman and Niklas,33 was also utilized. In the stochastic ENDOR sequence, RF is randomly hopped over the frequency range of the spectrum, with subtraction of a background signal (RF off) at each frequency. All measurements were done at 2 K. As desired, Q-band samples were photolyzed by placing them in liquid nitrogen in an X-band finger Dewar and illuminating them with a 1000 mW blue LED. The samples were then transferred to a liquid helium cryostat for EPR/ENDOR study.

Figure 4. X-band EPR spectra of MoFe protein (α-70Ile) freezetrapped during Ar turnover in H2O before (black) and during irradiation with 450 nm diode laser at 12 K (blue, 2.5 min, and red, 20 min traces). Red arrows highlight the conversion of E4(4H) to the photoinduced S state. Dashed green spectrum shows that annealing (Ann) the illuminated sample at 217 K for 2 min causes complete reversion of S to E4(4H). EPR conditions: T = 12 K; microwave frequency, 9.36 GHz; microwave power, 10 mW; modulation amplitude, 13 G; time constant, 160 ms; field sweep speed, 20 G/s.

freeze-trapped during turnover of MoFe protein that was remodeled with the α-70Val→Ile substitution; similar results are seen for MoFe with the double substitution, α-70Val→Ile/ α-195His→Gln (Figure 1). Both single and double substitutions prevent access of substrates other than protons, without perturbing FeMo-co function (Table S1), and enhance the accumulation of E4(4H).10−12 Figure 4 includes spectra collected during in situ photolysis with 450 nm light with the sample held at 12 K; equivalent spectra are obtained by photolysis at 77 K (e.g., Figure S2). Although it has not yet proven possible to create optically transparent freeze-quenched samples of this intermediate, the figures show that irradiation nonetheless causes the progressive loss of the E4(4H) signal, g = [2.15, 2.007, 1.965], and accompanying appearance of a new signal, S, with g = [2.098, 2.0, 1.956]. EPR spectra collected over temperatures between 12 and 50 K (Figure S1) show the E4(4H) signal disappears by ∼40−50 K because of rapid spin relaxation, whereas for S, the EPR signal is clearly visible at 50 K, demonstrating differences between excited spin-state manifolds of S and E4(4H). Careful examination of the timecourse of photolysis shows no new signals that are generated, other than S. In particular, Figures 4, S1, and S2 show no other photoinduced change in the spectrum in the vicinity of g ≈ 2; likewise, there is no change at lower fields (not shown), where the signal from residual resting state (E0) appears and where signals would appear from the state which has accumulated two [e−/H+] (E2(2H)).34 The photolysis product, S, completely relaxes back to E4(4H) when annealed for 2 min at 217 K, with no other change in the EPR spectrum, either in the g ≈ 2 region (Figure 4) or in the high-g (low-field) region (not shown). C

DOI: 10.1021/jacs.5b11650 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society Thus, photolysis/annealing cause the reversible conversion, E4(4H) ⇔ S. 1,2 H ENDOR spectroscopy was used to determine the fate of the [Fe−H−Fe] bridging hydrides during the photolytic E4(4H) ⇒ S conversion. Figure 5 shows 2 K Q-band stochastic

created by photolysis of E4(4D) prepared in D2O buffer show the loss of the [Fe-D-Fe] signals associated with E4(4D), again with no new strongly coupled signal appearing (not shown). The absence of new strongly coupled signals in the 1,2H ENDOR responses further rules out the possibility that S is an H2 complex of FeMo-co, either with strongly hindered or free rotation of H2, rather than having released H2. A bound H2 complex with strongly hindered rotation would show new strongly coupled 1H ENDOR signals, contrary to observation.35 Quantum statistical arguments show that freely rotating H2, which can occur even at 2 K, would not show 1H ENDOR signals, but would show new strongly coupled 2H ENDOR signals for the corresponding D2O sample.36 The absence of such new signals for S means that it does not contain a freely rotating bound D2. We thus infer that the state S generated by photolysis of E4(4H) has indeed lost the metal-bound hydrides through the release of one or two H2. Mechanism of Photoconversion. Three mechanisms must be considered for the E4(4H) ⇒ S photoconversion through loss of both hydrides and release of H2 (Figure 6),

Figure 5. Q-band stochastic 1H CW ENDOR spectra showing loss of signals from hydrides, H1 and H2, through photolysis. (Black) Before and (red) after 450 nm photolysis of MoFe protein (α-70Ile/α-195Gln) trapped during Ar turnover in H2O buffer. ENDOR conditions: microwave frequency, ∼34.99 GHz; modulation amplitude, 6.3 G; RF duration, 3 ms; RF cycle, 200 Hz; bandwidth of RF broadened to 100 kHz; 2000 scans; temperature, 2 K.

CW 1H ENDOR spectra collected at two different g-values from E4(4H) before and after photolysis; Figure S2 presents corresponding “conventional” CW spectra. In both figures, the spectra represent components of the 2D field-frequency pattern of spectra collected across the EPR envelope, which has been thoroughly analyzed in terms of two hydrides with anisotropic hyperfine tensor components that are virtually identical, but with tensors that are differently oriented with respect to g.10 In the g = 2.01 spectrum of E4(4H) in Figure 5, the strongly coupled 1H signals from the two hydrides (A ≲ 40 MHz) are completely overlapped; at g = 2.00, the signals again overlap, but distinct peaks also are seen from the individual hydrides, most noticeably the two doublets with couplings of A ≈ 22 and 28 MHz.10,27 For the sample that gave the ENDOR spectra in Figure 5, the photolysis had reduced the intensity of the E4(4H) EPR signal by ∼3-fold, with the corresponding appearance of EPR intensity from S. Figure 5 shows that the photolysis decreases the intensity of the 1H ENDOR signals from the strongly coupled hydrides by a comparable amount, without the appearance of new strongly coupled signals that can be associated with a metal-hydride “isomer”. At g = 2.01, no new signal appears; at g = 2.0, a signal appears with the relatively small coupling, A ≈ 10 MHz, which likely is associated with one of the protonated sulfides that must be present in S. CW ENDOR spectra taken on a sample with an even greater extent of photolysis show essentially complete loss of the ENDOR signals at a field where the two hydrides show distinct features (Figure S2). Correspondingly, 2H ENDOR measurements of S

Figure 6. Alternative mechanisms for the E4(4H) ⇒ S photoconversion through loss of both hydrides, release of H2, and thermal reverse.

which we denote re, hp, and L. The first would be precisely the photoinduced reductive-elimination of one H2, as discussed above, yielding a nonthermal form of the doubly reduced state of the cluster, which we denote E4(2H)*. The second postulates two steps of photoinduced hydride protonolysis (hp), each with loss of H2, in which case S must be a low-spin (S = 1/2) “spin-isomer” of the E0, S = 3/2 resting state, which we can denote E0†. Thermal hp processes are responsible for H2 formation by nitrogenase,1,8,9 as well as by hydrogenases.37,38 The third mechanism differs from the other two in that it does not involve generation of H2. Instead it would involve two steps of photoinduced re of a hydride, with transfer of the proton “released” to a bridging sulfide. We denote this the “L mechanism” because it is inspired by the photoinduced conversion of the Ni−C state of the Ni−Fe hydrogenases.37,38 Ni−C exhibits a [Ni(III)-H-Fe(II)] bridging hydride and a cysteinyl thiolate bound to Ni. Photolysis generates a state, denoted Ni-L, which contains [Ni(I), Fe(II)] metal ions, and with the proton formed by photoinduced reductive elimination of the hydride having been transferred to the bound sulfur. D

DOI: 10.1021/jacs.5b11650 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society Any one of these three imagined processes would yield a final photoproduct without metal hydrides. The observation that S thermally relaxes to E4(4H) immediately rules out the hp mechanism for photoinduced E4(4H) ⇒ S conversion. This conversion would involve the loss of four reducing equivalents as two H2 molecules, and its reverse must involve two steps of reduction of FeMo-co by H2: E0† + H2 ⇒ E2(2H); E2(2H) + H2 ⇒ E4(4H). However, it is one of the foundational facts about nitrogenase mechanism that H2 cannot react with any thermally generated state of FeMo-co except the N2-bound state produced by re of H2; as we have explained, the reactions are uphill by ∼30 kcal/mol.8 Even if the E0† photogenerated spin isomer were sufficiently activated as to react with H2, the first step of reduction would necessarily produce the thermally equilibrated E2(2H), which could not react with H2: hence E4(4H) could not be regenerated from S during cryoannealing. The complete absence during photolysis of a second new EPR signal in addition to S (see above) in fact not only independently rules out hp, but also rules out L. Both mechanisms involve sequential steps of photon absorption. Under the constant low-level illumination of this experiment this necessarily implies the buildup then loss of the EPR signal from the intermediate stage that has absorbed one photon. Thus, we conclude that photolysis indeed generates a FeMo-co state, S = E4(2H)*, through the photoinduced reductive elimination of the two bridging hydrides of E4(4H) with accompanying production of one H2, and that the relaxation of E4(2H)* to E4(4H) during cryoannealing corresponds to the oxidative addition of H2 to the photogenerated state, E4(2H)*. Kinetics of H2/D2 oa. To characterize the isotope dependence of the thermal oa process, E4(2H)* + H2 ⇒ E4(4H) versus E4(2D)* + D2 ⇒ E4(4D),29,30,39 we measured the kinetics of the 193 K relaxation of E4(2H)* and E4(2D)* in samples prepared, respectively, in H2O and D2O buffers. In these experiments the sample was annealed at 193 K for multiple time intervals, with cooling to 12 K for collection of EPR spectra between intervals. For both H2O and D2O buffers the relaxation is well-modeled as a one-step process (Figure 7). This is consistent with, but not proof of, the absence of an intermediate state(s), for example an H2 complex.40

With S in H2O buffer, the exponential decay time-constant is, τ = 10 min; decay is slowed in D2O buffer, to the decay time, τ = 55 min. The E4(4H) state recovers in synchrony with the loss of S, as the recovery also is exponential, and exhibits the same time-constants, τ = 9 min for H2O sample and τ = 47 min for D2O. This kinetic isotope effect during oa, KIE ≈ 5.4, is larger than typical for closed-shell monometallic complexes.41,42 Combined with a strong temperature dependence in the time constant (not shown), this KIE implies that oa of H2 involves traversal of an energy barrier associated with H2 binding and/or bond cleavage. The exponential decay of S and appearance of E4(4H) suggest that in the frozen solution the H2 formed by photoinduced re is trapped adjacent to FeMo-co and undergoes oa “intramolecularly”, presumably in some part because the incorporation of isoleucine over the active face of FeMo-co prevents H2 diffusion away. Indeed, although we presume that the relatively weakly coupled, but clearly resolved 1H ENDOR signal seen for S is associated with a sulfur-bound proton, we cannot rule out the possibility that it comes from the H2 trapped nearby. KIE of Photoinduced re. We know of only two studies of the KIE for photoinduced re of H2/D2, and these found small43 or negligible isotope effects.21 For completeness, we nonetheless used in situ photolysis to compare the time-dependent loss of the E4(4H) and corresponding E4(4D) signals as a function of temperature (Figure 8). In a clear solution of low optical

Figure 8. Time course of in situ 450 nm photoinduced conversion of E4(4H) intermediate trapped during MoFe protein (α-70Ile) turnover in H2O (lower) and D2O (upper). Photolysis at 3.8 (green), 8 (red), and 12 K (blue). Signal measured directly as intensity of the g1 feature of the E4(4H) S = 1/2 EPR signal, normalized to the maximum signal and fit with a stretched exponential decay function, I = exp(−[t/τ]n), with “1/e” time constant τ; 0 < n ≤ 1 equals unity for exponential decay and decreases with the spread of the distribution. Time constants for fits (white dashed lines) are given in figure; in all cases n ≈ 0.4 (see SI for details). EPR conditions: microwave frequency, 9.36 GHz; microwave power, 10 mW (1 mW for measurements at 3.8 K); modulation amplitude, 13 G; time constant, 160 ms.

density, photolysis under constant illumination would cause an exponential loss of the signal with rate constant (k) proportional to the light intensity (I0) and quantum yield (ϕ), for re, k(ϕ) = τ−1(ϕ) ∝ I0ϕ, where τ(ϕ) is the corresponding timeconstant for decay. However, it proved impossible to prepare clear frozen samples with high population of E4(4H), only frozen “snows”. The decay of the signal during photolysis of a “snow” is necessarily nonexponential because light scattering diminishes the photon flux across the sample (see SI); instead, as shown in Figure 8, it can be described with a stretched exponential, exp(−[t/τ]n),44 with ϕ-dependent “1/e” decay

Figure 7. Decay during 193 K annealing of E4(2H)* photoinduced in MoFe (α-70Ile) freeze-trapped during turnover in H2O (red) and D2O (blue), along with the parallel recoveries of E4(4H). Data points obtained as intensities of g1 feature of the corresponding S = 1/2 EPR signals, normalized to the maximum signal; they were fit with an exponential function, with time constants shown in the figure. EPR conditions: as in Figure 4. E

DOI: 10.1021/jacs.5b11650 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society time, τ(ϕ), whose inverse is proportional to the average decay rate constant (k̅)44 and is thus proportional to the quantum yield for re: τ−1(ϕ) ∝ k̅(ϕ) ∝ ϕ (see SI). The decay time for photoinduced re of dihydrogen from E4(4H) measured in both H2O and D2O buffers is temperature invariant, within error, from 4 to 12 K (Figure 8, Table S2), a variation in the thermal energy (kBT) by a factor of ∼3. Contrary to expectation, the decay slows markedly in D2O buffer; the KIE for re over this range, defined as the ratio of the “1/e” decay times for D2O and H2O buffers, is large, KIE ≈ 10. These observations together imply that photoinduced re involves a barrier to the combination of the two nascent H atoms, in contrast to the barrierless process inferred for monometallic metal complexes,17,21 and suggest that the photoinduced formation of H2 involves nuclear tunneling through that barrier. Whether the process involves an actual intermediate H2 complex remains to be determined. In combination with the evidence for a barrier crossing in the oa of H2 to E4(2H)* to regenerate E4(4H), this leads us to the picture of the energy surfaces for photoinduced re/oa for the Janus intermediate presented in Figure 9.

Figure 10. Cartoon showing nodal properties of an excited MO for an M(H)2 complex that is bonding between the two hydrides and antibonding between each one and the metal dz2 orbital.

be concerted, and to occur on a barrierless repulsive energy surface, with the H’s freely coming together for release as H2.17,21 This view is compatible with measurements showing prompt release of H2 after photolysis (